Fig 1.
Overview of C-ROADS model, adapted from Sterman et al. [27].
Participants in World Climate specify CO2 emissions from fossil fuels or land use, land use change, and forestry (LULUCF).
Fig 2.
Screenshots from the C-ROADS World Climate computer model.
Panel A is the six-region World Climate interface through which participants enter decisions, including the year they choose (if any) to halt the growth of emissions, begin to decline and the annual rate of decline (%), with changes in deforestation and afforestation (on scales of 0–100%, with 0 being business-as-usual and 100% being the maximum possible effort). The model immediately displays the resulting CO2 emissions trajectories (panel A left), global mean surface temperature anomaly relative to pre-industrial levels (panel A right) and other impacts. B: Screenshot showing CO2 emissions and net removals for the scenario entered in panel A (panel B left), illustrating the “carbon bathtub” [33], i.e., that the stock of CO2 in the atmosphere accumulates anthropogenic CO2 emissions less the net CO2 flux from the atmosphere to biosphere and oceans. Users can carry out a wide range of sensitivity tests by choosing values for parameters affecting, e.g., climate sensitivity and the strength of both positive (e.g., Arctic methane) and negative (e.g., CO2 fertilization) feedbacks in the climate system (panel B, bottom left). C: C-ROADS enables users to explore economic and population data linked to emissions (e.g., GHG emissions per capita shown in panel C, left), and to compare the fit between the model and historical data for GHG concentrations and to projected global surface temperature in CMIP5 models through 2100.
Fig 3.
Sequence of a World Climate simulation.
Fig 4.
Theoretical model of learning for action through the World Climate simulation.
Table 1.
Overview of World Climate sessions and participants in this study.
Table 2.
Participants and usable cases for each World Climate session.
The number of pre-, post-, and matched surveys obtained, expressed as a percentage of the total number of participants in a given session.
Table 3.
Factor loadings and communalities based on principal axis factor analysis with orthogonal rotation (varimax with Kaiser normalization) from pre-survey item analysis (N = 1,059; Kaiser-Meyer-Olkin measure of sampling adequacy = 0.89; Bartlett's test of sphericity p <1E-9).
Table 4.
Factor loadings and communalities based on principal axis factor analysis with orthogonal rotation (varimax with Kaiser normalization) from post-survey item analysis (N = 914; Kaiser-Meyer-Olkin measure of sampling adequacy = 0.89; Bartlett's test of sphericity p < 1E-9).
Table 5.
Comparison of pre- and post-survey means for constructs and survey items reflecting climate change knowledge (‘Impacts,’ ‘Causes’, ‘Stock-flow’), affect (‘Urgency,’ and ‘Hope’), and intent to take action (‘Intent’).
Table 6.
Analysis of gains and effect sizes for participants who began the simulation with low (lower third) vs. high (upper third) pre-survey values of each construct.
Fig 5.
Post-survey responses to questions regarding (A) how engaging the World Climate simulation was as a learning experience, (B) the effects the simulation had on motivation to address climate change and (C) desire to learn more about climate change science, solutions, politics, economics, and policies; N ≧ 839.
Fig 6.
Summary of regression results, showing statistically significant relationships (arrows) among gains in constructs, including affect (Urgency and Hope), knowledge about Impacts, Intent to act and Desire to Learn More.
Results are for Model 1 (no participant- or session-level fixed effects; values for Models 2–4 with different sets of controls are similar). Lines with arrows depict statistically significant relationships between independent and dependent variables, with standardized beta coefficients for each relationship shown. See S1 Table for detailed regression results. *** p < 1E-9 denotes statistical significance at α < 0.001 after Bonferroni correction. ** Beta coefficients were statistically significant at α < 0.01 after Bonferroni correction, with p < 1E-6.
Table 7.
Comparison of construct means for US-based participants who were somewhat or strongly opposed to free market regulation compared to those somewhat or strongly in favor of regulation, before and after World Climate.
Table 8.
Comparison of pre- and post-survey results for constructs reflecting climate change affect (‘Urgency,’ and ‘Hope’), knowledge (‘Impacts,’ ‘Causes,’ ‘Stock-Flow Understanding’), and intent to take action (‘Intent’) for participants in the US who responded “somewhat opposed” or “strongly opposed” when asked, “To what extent are you in favor of the government placing regulations on the free market?”